Online Credit Card Fraud Analytics Using Machine Learning Techniques

Abstract

Every year, companies and financial institutions worldwide lose billions due to credit card fraud. With the advancement in electronic commerce and communication technology and the development of modern technology, there has been an increase in the usage of credit cards and also the risks associated with it. Financial fraud detection systems are being implemented everywhere to minimize these losses. In the real world, fraudulent and real transactions are scattered all around and it is extremely difficult to distinguish between them. There is a class imbalance problem due to the minority of fraudulent transactions. The enormity of imbalanced data poses a challenge on how well Machine Learning techniques can efficiently detect fraud with a high prediction accuracy, and at the same time reduce misclassification costs. In this paper, experiments are performed to compare the most commonly used ML algorithms and investigate their performance when applied to a massive and highly imbalanced dataset. We compare six Supervised Machine learning techniques, namely, Logistic Regression, K-Nearest Neighbours, Naive Bayes, Decision Trees, Random Forest and Linear Support Vector Machine (SVM).